Friday, November 29, 2013

Are Humans Long Pig? or Monkey-Fish-Frog?

Does the creature on the right look more like a primate?  It isn't.  It is a non-primate mammal with a severe developmental anomaly, a mammal that would not ordinarily have a very primate-like face.  I don't think anyone, as yet, knows what causes this syndrome.  I don't think the syndrome has ever been considered important enough to investigate.

I came across a very interesting series of articles at macroevolution.net by a Dr. Eugene McCarthy with a PhD in genetics. He specializes in cross-species hybridization (he has a book on avian examples), and he makes a thought-provoking argument that because humans share so many traits uncharacteristic of the great apes and that many of these traits are similar to those  of a particular non-primate mammal that the traits are best explained by an extremely unlikely hybridization event.  It sounds more than a little “monkey-fish-frog” (or man-bear-pig, if you get that reference), but even if the conclusion seems absurd, the evidence he brings up is compelling, the unlikelihood of a successful hybrid stabilization event notwithstanding.

There are many things about this theory (along with the reactions to it) that are engaging.  One is the following angle:  How well does Dr. McCarthy’s evidence support considering the human genome as a intentional incorporation of specific pieces of non-primate DNA into the basic ape architecture in a way  that required specific knowledge of the architecture?  That is, how much does it resemble a deliberate engineering effort?  (Aside: Some groups might be more supportive of pig-chimp theory if it declared that "all men are pigs" at the exclusion of the female gender.)

With all the evidence that McCarthy trots out, it seems like a wonder that humans are considered so closely related to apes.  Think of the paleontological primacy of bone structure and how it is undermined by marsupial phylogeny.  Based on bone structure alone, it would be impossible to realize how different many marsupials are from their convergently evolved  placental counterparts.  Dr. McCarthy has begun to question the assumptions of paleontology (if it's really old it must be something like a lizard) and whether some dinosaurs should be considered to be much more like mammals than reptiles.  (And how long did it take for paleontology to accept the idea that some dinosaur creatures could be more avian than reptilian?)

It seems like more and more geneticists and evolutionary biologists have come out and admitted that the “98% similarity” metric between humans and chimps is fundamentally flawed and misleading. (Why do biologists keep repeating it then?)  If McCarthy’s statistics are right (and consistent), the difference in the estimated numbers of genes between humans (37,381) and chimps (32,887) is staggering.  So did proto-humans develop 14% more novel genes (in how much evolutionary time?) or were 4494 unneeded genes thrown away as chimps devolved?  
I don't think that McCarthy has produced any strong evidence that animals as dissimilar as apes and pigs can hybridize (other than trotting out humans themselves, which is as circular as many evolutionary arguments), but that isn't the biggest weakness of the theory.  What seems to discredit the theory most is that the proposed mechanism of back-crossing and stabilization seems far-fetched in the extreme.  P.Z. Myers (much as I dislike agreeing with him) brings up these important mechanism problems in his usual humble and inoffensive way (he calls McCarthy's hypothesis the "MFAP" theory -- "Monkey 'Fooled' with A Pig").

I do think that selectionist criticisms of pig-chimp theory seem like stones thrown from a glass house in that the likelihood of functional information from pigs surviving massive gene conversion (so massive that it would obscure our porcine origins) is infinitesimal.  One might say that the argument against useful information surviving gene conversion on this order is analogous to thermodynamic arguments against unlikely events.  But, to use the hyperselectionists' own arguments against them, however improbable it may be, here we are.  Or this argument:  astronomically improbable events happen all the time.  Or (when all else fails) this gem of a logical neutron bomb: with an innumerable number of parallel universes it was bound to happen.  (Yes, otherwise reputable scientists seem to trot out that self-defeating nuke on a regularly.)

Even if far-fetched, it is a thought-provoking point of view, not unlike considering what Lynn Margulis’ theories explain better than selectionism. And who knows? Since it is a materialist explanation, maybe biology will warm up to McCarthy’s theory in 20 years. There are discrepancies in the “molecular clock” of the X/Y chromosomes that have prompted biologists to hypothesize a cross-species rehybridization event between the hypothetical proto-humans and proto-chimps, so some unusual events are already on the table. The loss of vitamin C manufacture is another oddity pointing to an evolutionary “close call.”

Serendipitous vectoring of useful genes (and subsequent co-opting) from one organism to a very different organism may be unlikely, but horizontal gene transfer it seems is extremely common.  So common in fact that when biomolecular data is used to tighten up the phylogenetic relationships of mammals, it becomes a total mess.  And not just mammals, but the whole animal kingdom phylogeny seems to suffer from this problem:
Syvanen recently compared 2000 genes that are common to humans, frogs, sea squirts, sea urchins, fruit flies and nematodes. In theory, he should have been able to use the gene sequences to construct an evolutionary tree showing the relationships between the six animals. He failed. The problem was that different genes told contradictory evolutionary stories. This was especially true of sea-squirt genes. Conventionally, sea squirts—also known as tunicates—are lumped together with frogs, humans and other vertebrates in the phylum Chordata, but the genes were sending mixed signals. Some genes did indeed cluster within the chordates, but others indicated that tunicates should be placed with sea urchins, which aren't chordates. “Roughly 50 percent of its genes have one evolutionary history and 50 percent another,” Syvanen says. [quoted here from this]
Another interesting, if tangential, aspect that stood out to me is that some of P.Z. Myers’ criticisms seem to recognize the software-like aspects of incompatible genotype architectures.
Development is like a ballet, in which multiple players have to be in the right place and with the right timing for everything to come off smoothly. If someone is out of place by a few feet or premature by a few seconds in a leap, the dancers could probably compensate because there are understood rules for the general interactions…but it would probably come off as rough and poorly executed. A hybrid between two closely related species would be like mixing and matching the dancers from two different troupes to dance similar versions of Swan Lake — everything would be a bit off, but they could probably compensate and muddle through the performance. 
Hybridizing a pig and a chimp is like taking half the dancers from a performance of Swan Lake and the other half from a performance of Giselle and throwing them together on stage to assemble something. It’s going to be a catastrophe.  ["The MFAP Hypothesis"]
Of course, all of these difficulties are, in Myers' opinion, surmountable by the inherent gradualism of natural selection.  By proceeding gradually, natural selection can vault the following sorts of hurdles:
For example, we have bigger brains than chimpanzees do. This is not a change that was effected with a single switch; multiple genes had to co-evolve together, ratcheting up the size in relatively incremental steps. So you could imagine a change that increased mitotic activity in neural precursors that would increase the number of neurons, but then you’d also need changes in how those cells are partitioned into different regions, and changes in the proliferation of cartilage and bone to generate a larger cranium, and greater investment in vascular tissue to provide that brain with an adequate blood supply. [emphasis mine]
The holistic aspect of organism design (a concept that Elsberry and Shallit mock when it comes to the Intelligent Design hypothesis) routinely requires several tailored changes of separate parameters in the design.  This is a huge problem that is downplayed in any discussion of what natural selection must be able to handle.  And it is belittled precisely because it scandalously evokes ideas of "irreducible complexity."  Epistasis is an enormous complication for natural selection, and in the end, the evidence that natural selection is capable of regularly vaulting such epistatic hurdles is the very data that neo-Darwinism is supposed to explain: homologies and the fossil record.

This is a type of complexity that prompts information theorists to reach beyond Kolmogorov complexity to discuss "logically deep" and "computationally deep" bit sequences.  The sort of complexity that makes a novelist change a word on page 89 in order to fit with events and descriptions on pages 147 and 371.  The sort of complexity that makes an airplane manufacture overhaul the design of one component to allow the vehicle as a whole to meet a functional requirement.  Any bit string describing either of these artifacts would have dauntingly epistatic relationships between many bits.  Of course, if there is a lot of information front-loaded into the basic architectural of all animals, maybe Myers' dance coordination metaphor is overblown.  But in engineering, it is difficult to parameterize a design to allow such flexibility.  I'm sure the "evolution defenders" under the right circumstances, can marshal plenty of evidence, that organism designs are indeed very flexible.

Nontrivial history between genomes
P.Z. Myers seems to be describing the problem of mixing two very different "computationally deep" configurations, and implicitly arguing that a description of the cumulative differences between the two configurations would itself be more like computationally deep string than a merely random string.  A random mixing (well, semi-random: McCarthy's proposed back-crossing invokes natural selection, that praise-worthy almighty creative power) of two strings separated by a chasm of computational depth is much more likely to destroy functional information than to create it.
Rearranged software modules in pigs and apes.

Comparing sections of genetic code that vary greatly between hominids and pigs show a significant amount of rearrangement of sections of "code" (software).   Programmers who have re-factored and re-organized code may recognize the problem of creating difference reports after the software evolution exhibits "nontrivial history" between the versions of the program being compared.

What if there were evidence that a new vehicle incorporated technology from both a race car and an airplane in a novel way, a novel combination of features?  How difficult would it be to co-opt design features from one and adapt it to evolve the other for a new purpose?  Would it require some sort of intelligent design?

Maybe McCarthy is somewhat right about the evidence even though wrong about the mechanism.  Many animals herd other animals.  If apes herded porcine critters at some point, maybe there was a lot of viral vectoring of genetic data into their DNA.  If this seems far-fetched then you might want to rethink the whole concept of lateral/horizontal gene transfer (which is a catch-all name for absurdly improbable genetic similarities that contradict the supposed evolutionary history of the organism).  If “[r]oughly 50 percent of [sea squirt] genes have one evolutionary history and 50 percent another,” then what is the evolutionary history of the sea squirt?

It's as though, above the genus level, all critters are to some extent genetic chimera.  But the genes aren't just fortuitously exchanged between members of the animal kingdom.
Although Elysia chlorotica are unable to synthesize their own chloroplasts, the ability to maintain the chloroplasts acquired from Vaucheria litorea in a functional state indicates that Elysia chlorotica must possess photosynthesis-supporting genes within its own nuclear genome; most likely acquired through horizontal gene transfer.[5] Since chloroplast DNA alone encodes for just 10% of the proteins required for proper photosynthesis, scientists investigated the Elysia chlorotica genome for potential genes that could support chloroplast survival and photosynthesis. The researchers found a vital algal gene, psbO (a nuclear gene encoding for a manganese-stabilizing protein within the photosystem II complex[5]) in the sea slug's DNA, identical to the algal version. [from Wikipedia, emphasis mine]
In case you missed it, the implicit evidence of horizontal gene transfer is the implausibility* of a protein-coding sequence in the snail genome evolving convergently to the same sequence in the algal genome.


*Richard Dawkins argues that implausibility arguments are a hallmark of creationism.  Secular creationists are everywhere, so be on your guard.  

No comments:

Post a Comment